New preprint: Including minor alleles improves fluoroquinolone resistance prediction

Fluoroquinolones are used to treat both normal and drug resistant tuberculosis and therefore being able to work out if an infection is resistant or not to fluoroquinolones is very important. Sequencing the genome of an infection is increasingly used to rapidly return which antibiotics could be used to treat a patient with tuberculosis. Genetics-based approaches usually assume that any infection is homogenous which allows the variant caller to assume that any evidence of a minor alleles are due to sequencing error, allowing these to be filtered out.

The WHO catalogue of mutations conferring resistance to M. tuberculosis was published in 2021 and includes several mutations in the gyrA gene that confer resistance to both moxifloxacin and levofloxacin. Despite the molecular mechanism being thought to be understood the sensitivity of genetics-based resistance prediction was lower for the fluoroquinolones than rifampicin and isoniazid.

In this preprint Alice Brankin uses the large CRyPTIC dataset of M. tuberculosis to show that if two or more reads at a genome position support the existence of a known resistance-conferring mutation in gyrA, then calling that sample resistant improves the sensitivity of moxifloxacin resistance prediction from 85.4% to 94.0%, bringing it into line with rifamipcin and isoniazid.